The past decade has seen the rise of voice assistants, such as Apple’s Siri, Amazon’s Alexa, and Google’s Google Assistant. On top of performing tasks such as booking a flight, turning the light on and off, and ordering a taxi, these voice assistants are also used to browse on the internet. However, this new way of accessing the World Wide Web brings its share of new concerns and limitations.
Voice assistants do raise questions about the use of artificial intelligence, surveillance capitalism, privacy, speech being used to activate and use voice assistants. However, this is not the focus of this research.
First, voice assistants are considered interfaces, which are defined as “devices or programs that allow users to interact with computer resources.” Voice assistants are based on software processing voice inputs, and therefore users cannot browse the internet as they would facing a computer screen. When using one keyword in a search engine like Google, users have access to a plurality of sources and results. However, using a voice assistant like Google Assistant, users will receive a limited amount of results, if not only one result. The reason for that is mostly practicality, but users do lose a lot of control when using voice assistants as search engines compared to traditional computers. In May 2020, the Web counted about 1.7 billion websites, and voice assistants present their users with only a percentage of that.
Secondly, voice assistants can seem to be appropriating knowledge and resources. The study gives the example of Wikipedia, used by voice assistants to answer user queries. Most of Wikipedia’s articles are redacted by volunteers, and when using knowledge created by Wikipedia, users get the perception that the material comes from voice assistants and not the volunteer editors, whose work gets “appropriated.”
Thirdly, voice assistants present a series of biases. Web interfaces are not neutral, since they are man-made devices. Results are ranked by algorithms that include location, language, and previous searches. Moreover, these assistants contain a number of assumptions about what is normal, valuable, and appropriate, notions that have been determined by designers and programmers who encode such assumptions into the devices. Voice assistants also make use of a particular voice, often human and gendered, associated with a name (Alexa, Siri, Cortana…) that can create stereotypes. Besides a gender (often female), the voice can also be perceived as associated with a particular race (often white, native speaker).
Finally, voice assistants raise questions regarding accessibility. The study states that “because the availability of large amounts of data on users is crucial to the development of effective voice processing systems that automatically transcribe voice inputs, it is more difficult to develop voice assistants for languages spoken by minority groups or in smaller communities.” Voice assistants are therefore not accessible in some languages. The same issue applies to the ability of small companies to develop their own voice assistant since the technologies and resources required for the development of such systems are not always accessible to private developers or small organizations outside of the main actors of digital capitalism (Google, Amazon, Apple…). There are also not accessible to people who do not dispose of an internet connection, which is necessary for the device to retrieve information and access resources. As for people with a disability such as speech or hearing impairment, these interfaces present difficulties regarding their use, hence the need for a plurality of interfaces besides the voice.
To read more: https://journals.sagepub.com/doi/10.1177/0163443720983295
Natale S, Cooke H. Browsing with Alexa: Interrogating the impact of voice assistants as web interfaces. Media, Culture & Society. 2021;43(6):1000-1016. doi:10.1177/0163443720983295